Greenhouse gas emissions, such as carbon dioxide, contribute to global warming (Malik et al., 2016), which is a threat for humans (IPCC, 2013; Ma et al., 2019). Studies have shown that urban areas account for 2% of the total land area, but produce 80% of the greenhouse gas emissions (Lombardi et al., 2017; Balk et al., 2005). In 2014, the urban population exceeded 65% of the global population (The United Nations, 2014). Cities are the main areas of human activity and greenhouse gas emissions (Li et al., 2018). To cope with global climate change and protect the environment, greenhouse gas emissions have drawn increasing attention (Wang et al., 2017a). However, urban expansion and population growth have driven increasingly intensified carbon dioxide emissions (CEs) in cities, particularly in China (Zha et al., 2010). In recent decades, China has experienced rapid economic growth, environmental damage, and massive energy consumption. In 2014, the CEs of China were 10.29 billion tons, 4.21 times that in 1990, far higher than those of the United States (5.25 billion tons) and other developed countries (https://data.worldbank.org).
To date, extensive validation of CEs and their environmental effects have been conducted (Baumert et al., 2019; Hamilton and Friess, 2018), and measuring CEs from energy consumption has received increasing attention (Chen et al., 2017; Gudipudi et al., 2016; Zhang et al., 2019). For example, Lu (2018) quantified and analyzed the impact of CEs from fossil fuel consumption at the national level. Wang (Wang et al., 2015) analyzed the driving factors of CEs in China. Sufficient research has been conducted on carbon dioxide accounting. Methods of measuring CEs from various processes are developed. Three methods are widely used for carbon accounting: bottom-up, top-down, and hybrid approaches (Gusti and Jonas, 2010; Deng et al., 2011; Andersen et al., 2019). The top-down method is mainly a production-based accounting approach and is extremely useful for quantifying energy flows between regions and countries (Lombardi et al., 2017). Although the top-down method requires less time and labor, it is not suitable for carbon accounting of small systems. The bottom-up method is used for a consumption-based accounting method, which is based on resource consumption at each step of human activity and life cycle analysis (LCA) (Lombardi et al., 2017). The bottom-up method has a relatively high accuracy, but time-consuming and requires a large amount of data (Lombardi et al., 2017). The hybrid method includes the synthesis of bottom-up, top-down, and other methods, such as environmental input-output analysis (EIOA)/LCA (Kjaer et al., 2015). These integrated approaches are suitable for carbon dioxide accounting when detailed energy data are not available.
Although there is extensive research on CEs, the methods are mainly focused on accounting and analyze the relationship between CEs and socio-economic. However, Quantification of fossil fuel carbon dioxide emissions (CEs) at fine space and time resolution is a critical need in climate change research and carbon cycle (Gurney et al., 2009; Crippa et al., 2020). Moreover, quantifying the changes in spatiotemporal patterns of urban CEs is important to understand carbon cycle and development carbon reduction strategies (Chuai and Feng, 2019). More recently, mapping CEs has gained attention among researchers. With the development of remote sensing technology, carbon monitoring satellites have realized the dynamic monitoring of large-scale greenhouse gas emissions (e.g., greenhouse gas observation satellites, GOSAT). GOSAT's sub-satellite spatial resolution is approximately 10.5 km, and it is suitable for monitoring carbon dioxide concentrations on region and global. However, GOSAT images are not suitable for studying urban-scale CEs. Additionally, nighttime lighting (NTL) data sets are widely used to map CEs (Wang and Li, 2017; Su et al., 2014; Racitia et al., 2014; Zhang, et al., 2013), as there is a close relationship between NTL and population density, where a higher NTL value indicates greater energy consumption (Sutton et al., 2003; Zhuo et al., 2009). The spatial CEs distribution of existing studies has relied on spatial proxies to downscale the total CEs to a grid. Linear regression and panel regression models are the two most commonly used methods (Shi et al., 2016; Cui et al., 2019; Han et al., 2018). The NTL downscale method mapping CEs has a resolution of 1 km, the CEs distribution lack of more spatial details and unable to distinguish industrial, tertiary, and residential CEs. Moreover, Cai (2018) developed the China high-resolution emission database (CHRED), which was constructed using the bottom-up method and a spatial resolution of 1 km, combined socio-economic data. Most of the existing spatial CEs studies were mapped based on the NTL at a spatial resolution of 1 km. Thus, NTL-based CEs distribution still has limitations. On one hand, the distribution of CEs does not reflect further details of its spatial heterogeneity at the city level. Low spatial resolution cannot satisfy the application of fine urban carbon management and emission reduction strategy development. On the other hand, the downscaling method of CEs does not distinguish sectors CEs. To improve the spatial resolution of CEs, researchers have set up monitoring sites to accurately monitor CEs (Cai et al., 2019). In recent years, emerging big data has provided an opportunity to study the spatial characteristics of high-resolution urban CEs. Web crawler technology (WCT) can be used to obtain information about the location and attributes of the CEs. In addition, points of interest (POI), which contain multi-attribute information and location, is an important data source. For example, Gao(2021) used POI to mapping urban carbon emissions. Power point and industrial facilities were used to map CEs at high resolutions (Cai et al., 2014). In summary, existing CEs distribution research focus on region and global, based on NTL with 1 km resolution. So, urban and fine resolution CEs data is still rare.
In order to meet the critical needs of carbon cycle, urban precision carbon reduction strategies and carbon management, in this study, with the help of POI data, we developed an improved method and mapped urban CEs with a fine resolution. Compare to traditional NTL-based method, the CEs distribution of this study has better accuracy and fine resolution. Moreover, different sectors CEs distribution is clear. We selected the international city of Shanghai as an example to map the CEs and validate the method. This study used POI data and GIS technology to map the fine resolution urban CEs distribution. First, according to data availability and accuracy requirements, this study applied the bottom-up method to account urban CEs for residential, tertiary, and industrial sectors from 2000 to 2015. Moreover, this study used the POI data and multi sources data which obtained by web crawler technology (WCT). The method of mapping CEs in this study could clearly identify the CEs spatial patterns of each sector, and different from existing studies. We analyzed the spatial patterns of CE in Shanghai. We believe that this methodology could be applied in other fast-growing cities to understand carbon cycle and develop accuracy urban carbon reduction strategies (Wang et al., 2014).